1. Field of the Invention
The present invention generally relates to an apparatus and method for recognizing characters, and more particularly, to an apparatus and method for recognizing characters whereby a device having a photographing function recognizes a character included in a captured image and re-recognizes the character that is erroneously recognized after correcting the recognized character according to a user input.
2. Description of the Related Art
Existing character input methods used in mobile devices such as mobile phones include keypad input, character touch input, and handwriting input. Studies have been conducted in order for a mobile phone having a camera function to directly capture an image including a character of a direction board of a street or a character of a menu board of a restaurant, recognize the character, and provide information about the recognized character to a user, thereby allowing the user to know the meaning of the character even when the character is expressed in unfamiliar language.
In the case of handwriting input from among the character input methods, there are many variations in handwriting because the style and size of handwriting are different from user to user according to users' unique handwriting habits. As a result, it is difficult to guarantee stable recognition performance. In particular, a user has a difficulty in correctly inputting other unfamiliar languages, e.g., English, Chinese, Arabic, and Hebrew, than the native language, thus leading to further degradation in the performance of recognition of a user's handwriting input. In this regard, the performance of voice recognition of English words spoken by an English-speaking user degrades much more than the performance of voice recognition of English words spoken by a non-English speaking user.
On the other hand, recognition of a character included in an image captured by a camera is very useful in that such recognition does not require a user's keypad input or handwriting input. Character images can be easily obtained by a camera of a mobile device regardless of a medium on which target characters are recorded, such as a notice in a tourist resort, a signboard of a store, a menu board, or a name card. When a character included in a captured image is a print character, the character is less affected by influences from variations in handwriting input. However, good recognition performance cannot be guaranteed for a character in a graphic style, and since an environment where an image including a character is captured is not limited, a dull image may be obtained due to surrounding lighting or a photographing position or a blurry image may be obtained due to inaccurate focus control. Moreover, when character recognition is not performed accurately, the user has to experience the inconvenience of capturing an image including the same character again.
In the case of keypad input, the user cannot input an unfamiliar language or a non-native language for which input is not supported by a mobile device. Even when the mobile device supports input for such a language, the user may have a difficulty in thoroughly knowing a character input method for the unfamiliar language or the non-native language. In the case of handwriting input, recognition performance may degrade because a user's handwriting input for the unfamiliar language or the non-native language is not natural. Moreover, since additional image processing of handwriting input requires a large amount of computation, handwriting input cannot be actually applied to a mobile device having a cheap processor mounted therein.